Development of a risk score predicting survival after adult heart transplantation in the United States.

BACKGROUND

In the US, donor hearts for transplant are currently allocated to the candidates with the highest risk of death on the waiting list, based on a categorical status-based system. The upcoming continuous distribution system provides an opportunity to implement a post-transplant risk score that may help avoid futile transplants.

METHODS

In this observational study of the Scientific Registry of Transplant Recipients (SRTR), a novel US transplant risk score (US-TRS) was developed and validated using a mixed-effects Cox proportional hazards model. Study participants included adult heart transplant recipients between October 18, 2018, and February 28, 2022, split temporally into training (first 70% of recipients) and test (last 30% of recipients) datasets. We included 8 of 9 French Transplant Risk Score (French-TRS) components plus additional US variables that improved the Akaike Information Criterion (AIC) in the training data. In the test dataset, we assessed US-TRS 1-year post-transplant survival predictions with Uno's concordance index (c-index) and restricted mean survival time (RMST) and compared these results to other post-transplant risk scores, including the French-TRS, the Index for Mortality Prediction After Cardiac Transplantation (IMPACT) score, and recipients' waitlist status at transplantation.

RESULTS

The study cohort consisted of 9,071 heart transplant recipients (mean age 54 [SD 13] years, 72% male), of which 828 (9.1%) died and 29 (0.3%) underwent retransplant within 1 year of transplant. The final US-TRS model included recipient age, bilirubin, estimated glomerular filtration rate (eGFR), albumin, durable left ventricular assist device (LVAD), diabetes, mechanical ventilation, congenital heart disease, donor age, donor sex, and donor-recipient size mismatch. The c-index in the test dataset was 0.671 (95% CI, 0.665-0.687) for the US-TRS model, 0.620 (95% CI, 0.611-0.632) for the French-TRS model, 0.598 for the calculated IMPACT score, and 0.551 (95% CI, 0.540-0.576) for waitlist Status at transplant. US-TRS estimated excellent survival for most recipients, but a minority (17%) of recipients were high risk with an estimated 1-year survival probability of 78% and an average estimated RMST of 311.9 days in the first year.

CONCLUSIONS

In this registry-based study of US adult heart transplant recipients, a multivariable risk score outperformed existing models in predicting 1-year post-transplant survival and may be useful for integrating post-transplant survival into the upcoming continuous distribution framework.